* [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB
@ 2020-09-18 10:30 Ard Biesheuvel
2020-09-18 10:30 ` [RFC/RFT PATCH 1/6] ARM: p2v: factor out shared loop processing Ard Biesheuvel
` (7 more replies)
0 siblings, 8 replies; 14+ messages in thread
From: Ard Biesheuvel @ 2020-09-18 10:30 UTC (permalink / raw)
To: linux-arm-kernel
Cc: linux-efi, Nicolas Pitre, Linus Walleij, Russell King,
Santosh Shilimkar, Zhen Lei, Ard Biesheuvel
This series is inspired by Zhei Len's series [0], which updates the
ARM p2v patching code to optionally support p2v relative alignments
of as little as 64 KiB.
Reducing this alignment is necessary for some specific Huawei boards,
but given that reducing this minimum alignment will make the boot
sequence more robust for all platforms, especially EFI boot, which
no longer relies on the 128 MB masking of the decompressor load address,
but uses firmware memory allocation routines to find a suitable spot
for the decompressed kernel.
This series is not based on Zhei Len's code, but addresses the same
problem, and takes some feedback given in the review into account:
- use of a MOVW instruction to avoid two adds/adcs sequences when dealing
with the carry on LPAE
- add support for Thumb2 kernels as well
- make the change unconditional - it will bit rot otherwise, and has value
for other platforms as well.
The first four patches are general cleanup and preparatory changes.
Patch #5 implements the switch to a MOVW instruction without changing
the minimum alignment.
Patch #6 reduces the minimum alignment to 2 MiB.
Tested on QEMU in ARM/!LPAE, ARM/LPAE, Thumb2/!LPAE and Thumb2/LPAE modes.
Cc: Zhen Lei <thunder.leizhen@huawei.com>
Cc: Russell King <rmk+kernel@armlinux.org.uk>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Nicolas Pitre <nico@fluxnic.net>
[0] https://lore.kernel.org/linux-arm-kernel/20200915015204.2971-1-thunder.leizhen@huawei.com/
Ard Biesheuvel (6):
ARM: p2v: factor out shared loop processing
ARM: p2v: factor out BE8 handling
ARM: p2v: drop redundant 'type' argument from __pv_stub
ARM: p2v: use relative references in patch site arrays
ARM: p2v: switch to MOVW for Thumb2 and ARM/LPAE
ARM: p2v: reduce p2v alignment requirement to 2 MiB
arch/arm/Kconfig | 2 +-
arch/arm/include/asm/memory.h | 58 ++++++---
arch/arm/kernel/head.S | 136 ++++++++++++--------
3 files changed, 123 insertions(+), 73 deletions(-)
--
2.17.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 14+ messages in thread
* [RFC/RFT PATCH 1/6] ARM: p2v: factor out shared loop processing
2020-09-18 10:30 [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB Ard Biesheuvel
@ 2020-09-18 10:30 ` Ard Biesheuvel
2020-09-18 10:30 ` [RFC/RFT PATCH 2/6] ARM: p2v: factor out BE8 handling Ard Biesheuvel
` (6 subsequent siblings)
7 siblings, 0 replies; 14+ messages in thread
From: Ard Biesheuvel @ 2020-09-18 10:30 UTC (permalink / raw)
To: linux-arm-kernel
Cc: linux-efi, Nicolas Pitre, Linus Walleij, Russell King,
Santosh Shilimkar, Zhen Lei, Ard Biesheuvel
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
arch/arm/kernel/head.S | 24 +++++++++-----------
1 file changed, 11 insertions(+), 13 deletions(-)
diff --git a/arch/arm/kernel/head.S b/arch/arm/kernel/head.S
index f8904227e7fd..9a0c11ac8281 100644
--- a/arch/arm/kernel/head.S
+++ b/arch/arm/kernel/head.S
@@ -641,7 +641,7 @@ __fixup_a_pv_table:
#ifdef CONFIG_THUMB2_KERNEL
moveq r0, #0x200000 @ set bit 21, mov to mvn instruction
lsls r6, #24
- beq 2f
+ beq .Lnext
clz r7, r6
lsr r6, #24
lsl r6, r7
@@ -650,8 +650,8 @@ __fixup_a_pv_table:
orrcs r6, #0x0080
orr r6, r6, r7, lsl #12
orr r6, #0x4000
- b 2f
-1: add r7, r3
+ b .Lnext
+.Lloop: add r7, r3
ldrh ip, [r7, #2]
ARM_BE8(rev16 ip, ip)
tst ip, #0x4000
@@ -660,25 +660,21 @@ ARM_BE8(rev16 ip, ip)
orreq ip, r0 @ mask in offset bits 7-0
ARM_BE8(rev16 ip, ip)
strh ip, [r7, #2]
- bne 2f
+ bne .Lnext
ldrh ip, [r7]
ARM_BE8(rev16 ip, ip)
bic ip, #0x20
orr ip, ip, r0, lsr #16
ARM_BE8(rev16 ip, ip)
strh ip, [r7]
-2: cmp r4, r5
- ldrcc r7, [r4], #4 @ use branch for delay slot
- bcc 1b
- bx lr
#else
#ifdef CONFIG_CPU_ENDIAN_BE8
moveq r0, #0x00004000 @ set bit 22, mov to mvn instruction
#else
moveq r0, #0x400000 @ set bit 22, mov to mvn instruction
#endif
- b 2f
-1: ldr ip, [r7, r3]
+ b .Lnext
+.Lloop: ldr ip, [r7, r3]
#ifdef CONFIG_CPU_ENDIAN_BE8
@ in BE8, we load data in BE, but instructions still in LE
bic ip, ip, #0xff000000
@@ -694,11 +690,13 @@ ARM_BE8(rev16 ip, ip)
orreq ip, ip, r0 @ mask in offset bits 7-0
#endif
str ip, [r7, r3]
-2: cmp r4, r5
+#endif
+
+.Lnext:
+ cmp r4, r5
ldrcc r7, [r4], #4 @ use branch for delay slot
- bcc 1b
+ bcc .Lloop
ret lr
-#endif
ENDPROC(__fixup_a_pv_table)
.align
--
2.17.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [RFC/RFT PATCH 2/6] ARM: p2v: factor out BE8 handling
2020-09-18 10:30 [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB Ard Biesheuvel
2020-09-18 10:30 ` [RFC/RFT PATCH 1/6] ARM: p2v: factor out shared loop processing Ard Biesheuvel
@ 2020-09-18 10:30 ` Ard Biesheuvel
2020-09-18 10:30 ` [RFC/RFT PATCH 3/6] ARM: p2v: drop redundant 'type' argument from __pv_stub Ard Biesheuvel
` (5 subsequent siblings)
7 siblings, 0 replies; 14+ messages in thread
From: Ard Biesheuvel @ 2020-09-18 10:30 UTC (permalink / raw)
To: linux-arm-kernel
Cc: linux-efi, Nicolas Pitre, Linus Walleij, Russell King,
Santosh Shilimkar, Zhen Lei, Ard Biesheuvel
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
arch/arm/kernel/head.S | 30 +++++++++-----------
1 file changed, 14 insertions(+), 16 deletions(-)
diff --git a/arch/arm/kernel/head.S b/arch/arm/kernel/head.S
index 9a0c11ac8281..c2a912121e3e 100644
--- a/arch/arm/kernel/head.S
+++ b/arch/arm/kernel/head.S
@@ -669,26 +669,24 @@ ARM_BE8(rev16 ip, ip)
strh ip, [r7]
#else
#ifdef CONFIG_CPU_ENDIAN_BE8
- moveq r0, #0x00004000 @ set bit 22, mov to mvn instruction
+@ in BE8, we load data in BE, but instructions still in LE
+#define PV_BIT22 0x00004000
+#define PV_IMM8_MASK 0xff000000
+#define PV_ROT_MASK 0x000f0000
#else
- moveq r0, #0x400000 @ set bit 22, mov to mvn instruction
+#define PV_BIT22 0x00400000
+#define PV_IMM8_MASK 0x000000ff
+#define PV_ROT_MASK 0xf00
#endif
+
+ moveq r0, #PV_BIT22 @ set bit 22, mov to mvn instruction
b .Lnext
.Lloop: ldr ip, [r7, r3]
-#ifdef CONFIG_CPU_ENDIAN_BE8
- @ in BE8, we load data in BE, but instructions still in LE
- bic ip, ip, #0xff000000
- tst ip, #0x000f0000 @ check the rotation field
- orrne ip, ip, r6, lsl #24 @ mask in offset bits 31-24
- biceq ip, ip, #0x00004000 @ clear bit 22
- orreq ip, ip, r0 @ mask in offset bits 7-0
-#else
- bic ip, ip, #0x000000ff
- tst ip, #0xf00 @ check the rotation field
- orrne ip, ip, r6 @ mask in offset bits 31-24
- biceq ip, ip, #0x400000 @ clear bit 22
- orreq ip, ip, r0 @ mask in offset bits 7-0
-#endif
+ bic ip, ip, #PV_IMM8_MASK
+ tst ip, #PV_ROT_MASK @ check the rotation field
+ orrne ip, ip, r6 ARM_BE8(, lsl #24) @ mask in offset bits 31-24
+ biceq ip, ip, #PV_BIT22 @ clear bit 22
+ orreq ip, ip, r0 @ mask in offset bits 7-0
str ip, [r7, r3]
#endif
--
2.17.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [RFC/RFT PATCH 3/6] ARM: p2v: drop redundant 'type' argument from __pv_stub
2020-09-18 10:30 [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB Ard Biesheuvel
2020-09-18 10:30 ` [RFC/RFT PATCH 1/6] ARM: p2v: factor out shared loop processing Ard Biesheuvel
2020-09-18 10:30 ` [RFC/RFT PATCH 2/6] ARM: p2v: factor out BE8 handling Ard Biesheuvel
@ 2020-09-18 10:30 ` Ard Biesheuvel
2020-09-18 10:31 ` [RFC/RFT PATCH 4/6] ARM: p2v: use relative references in patch site arrays Ard Biesheuvel
` (4 subsequent siblings)
7 siblings, 0 replies; 14+ messages in thread
From: Ard Biesheuvel @ 2020-09-18 10:30 UTC (permalink / raw)
To: linux-arm-kernel
Cc: linux-efi, Nicolas Pitre, Linus Walleij, Russell King,
Santosh Shilimkar, Zhen Lei, Ard Biesheuvel
We always pass the same value for 'type' so pull it into the __pv_stub
macro itself.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
arch/arm/include/asm/memory.h | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/arch/arm/include/asm/memory.h b/arch/arm/include/asm/memory.h
index 99035b5891ef..eb3c8e6e960a 100644
--- a/arch/arm/include/asm/memory.h
+++ b/arch/arm/include/asm/memory.h
@@ -183,14 +183,14 @@ extern const void *__pv_table_begin, *__pv_table_end;
#define PHYS_OFFSET ((phys_addr_t)__pv_phys_pfn_offset << PAGE_SHIFT)
#define PHYS_PFN_OFFSET (__pv_phys_pfn_offset)
-#define __pv_stub(from,to,instr,type) \
+#define __pv_stub(from,to,instr) \
__asm__("@ __pv_stub\n" \
"1: " instr " %0, %1, %2\n" \
" .pushsection .pv_table,\"a\"\n" \
" .long 1b\n" \
" .popsection\n" \
: "=r" (to) \
- : "r" (from), "I" (type))
+ : "r" (from), "I" (__PV_BITS_31_24))
#define __pv_stub_mov_hi(t) \
__asm__ volatile("@ __pv_stub_mov\n" \
@@ -217,7 +217,7 @@ static inline phys_addr_t __virt_to_phys_nodebug(unsigned long x)
phys_addr_t t;
if (sizeof(phys_addr_t) == 4) {
- __pv_stub(x, t, "add", __PV_BITS_31_24);
+ __pv_stub(x, t, "add");
} else {
__pv_stub_mov_hi(t);
__pv_add_carry_stub(x, t);
@@ -235,7 +235,7 @@ static inline unsigned long __phys_to_virt(phys_addr_t x)
* assembler expression receives 32 bit argument
* in place where 'r' 32 bit operand is expected.
*/
- __pv_stub((unsigned long) x, t, "sub", __PV_BITS_31_24);
+ __pv_stub((unsigned long) x, t, "sub");
return t;
}
--
2.17.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [RFC/RFT PATCH 4/6] ARM: p2v: use relative references in patch site arrays
2020-09-18 10:30 [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB Ard Biesheuvel
` (2 preceding siblings ...)
2020-09-18 10:30 ` [RFC/RFT PATCH 3/6] ARM: p2v: drop redundant 'type' argument from __pv_stub Ard Biesheuvel
@ 2020-09-18 10:31 ` Ard Biesheuvel
2020-09-18 10:31 ` [RFC/RFT PATCH 5/6] ARM: p2v: switch to MOVW for Thumb2 and ARM/LPAE Ard Biesheuvel
` (3 subsequent siblings)
7 siblings, 0 replies; 14+ messages in thread
From: Ard Biesheuvel @ 2020-09-18 10:31 UTC (permalink / raw)
To: linux-arm-kernel
Cc: linux-efi, Nicolas Pitre, Linus Walleij, Russell King,
Santosh Shilimkar, Zhen Lei, Ard Biesheuvel
Free up a register in the p2v patching code by switching to relative
references, which don't require keeping the phys-to-virt displacement
live in a register.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
arch/arm/include/asm/memory.h | 6 +++---
arch/arm/kernel/head.S | 15 ++++++++-------
2 files changed, 11 insertions(+), 10 deletions(-)
diff --git a/arch/arm/include/asm/memory.h b/arch/arm/include/asm/memory.h
index eb3c8e6e960a..4121662dea5a 100644
--- a/arch/arm/include/asm/memory.h
+++ b/arch/arm/include/asm/memory.h
@@ -187,7 +187,7 @@ extern const void *__pv_table_begin, *__pv_table_end;
__asm__("@ __pv_stub\n" \
"1: " instr " %0, %1, %2\n" \
" .pushsection .pv_table,\"a\"\n" \
- " .long 1b\n" \
+ " .long 1b - .\n" \
" .popsection\n" \
: "=r" (to) \
: "r" (from), "I" (__PV_BITS_31_24))
@@ -196,7 +196,7 @@ extern const void *__pv_table_begin, *__pv_table_end;
__asm__ volatile("@ __pv_stub_mov\n" \
"1: mov %R0, %1\n" \
" .pushsection .pv_table,\"a\"\n" \
- " .long 1b\n" \
+ " .long 1b - .\n" \
" .popsection\n" \
: "=r" (t) \
: "I" (__PV_BITS_7_0))
@@ -206,7 +206,7 @@ extern const void *__pv_table_begin, *__pv_table_end;
"1: adds %Q0, %1, %2\n" \
" adc %R0, %R0, #0\n" \
" .pushsection .pv_table,\"a\"\n" \
- " .long 1b\n" \
+ " .long 1b - .\n" \
" .popsection\n" \
: "+r" (y) \
: "r" (x), "I" (__PV_BITS_31_24) \
diff --git a/arch/arm/kernel/head.S b/arch/arm/kernel/head.S
index c2a912121e3e..d2bd3b258386 100644
--- a/arch/arm/kernel/head.S
+++ b/arch/arm/kernel/head.S
@@ -633,7 +633,7 @@ ENDPROC(__fixup_pv_table)
__fixup_a_pv_table:
adr r0, 3f
ldr r6, [r0]
- add r6, r6, r3
+ add r6, r6, r0
ldr r0, [r6, #HIGH_OFFSET] @ pv_offset high word
ldr r6, [r6, #LOW_OFFSET] @ pv_offset low word
mov r6, r6, lsr #24
@@ -651,7 +651,8 @@ __fixup_a_pv_table:
orr r6, r6, r7, lsl #12
orr r6, #0x4000
b .Lnext
-.Lloop: add r7, r3
+.Lloop: add r7, r4
+ add r4, #4
ldrh ip, [r7, #2]
ARM_BE8(rev16 ip, ip)
tst ip, #0x4000
@@ -681,28 +682,28 @@ ARM_BE8(rev16 ip, ip)
moveq r0, #PV_BIT22 @ set bit 22, mov to mvn instruction
b .Lnext
-.Lloop: ldr ip, [r7, r3]
+.Lloop: ldr ip, [r7, r4]
bic ip, ip, #PV_IMM8_MASK
tst ip, #PV_ROT_MASK @ check the rotation field
orrne ip, ip, r6 ARM_BE8(, lsl #24) @ mask in offset bits 31-24
biceq ip, ip, #PV_BIT22 @ clear bit 22
orreq ip, ip, r0 @ mask in offset bits 7-0
- str ip, [r7, r3]
+ str ip, [r7, r4]
+ add r4, r4, #4
#endif
.Lnext:
cmp r4, r5
- ldrcc r7, [r4], #4 @ use branch for delay slot
+ ldrcc r7, [r4] @ use branch for delay slot
bcc .Lloop
ret lr
ENDPROC(__fixup_a_pv_table)
.align
-3: .long __pv_offset
+3: .long __pv_offset - .
ENTRY(fixup_pv_table)
stmfd sp!, {r4 - r7, lr}
- mov r3, #0 @ no offset
mov r4, r0 @ r0 = table start
add r5, r0, r1 @ r1 = table size
bl __fixup_a_pv_table
--
2.17.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [RFC/RFT PATCH 5/6] ARM: p2v: switch to MOVW for Thumb2 and ARM/LPAE
2020-09-18 10:30 [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB Ard Biesheuvel
` (3 preceding siblings ...)
2020-09-18 10:31 ` [RFC/RFT PATCH 4/6] ARM: p2v: use relative references in patch site arrays Ard Biesheuvel
@ 2020-09-18 10:31 ` Ard Biesheuvel
2020-09-18 10:31 ` [RFC/RFT PATCH 6/6] ARM: p2v: reduce p2v alignment requirement to 2 MiB Ard Biesheuvel
` (2 subsequent siblings)
7 siblings, 0 replies; 14+ messages in thread
From: Ard Biesheuvel @ 2020-09-18 10:31 UTC (permalink / raw)
To: linux-arm-kernel
Cc: linux-efi, Nicolas Pitre, Linus Walleij, Russell King,
Santosh Shilimkar, Zhen Lei, Ard Biesheuvel
In preparation for reducing the phys-to-virt minimum relative alignment
from 16 MiB to 2 MiB, switch to patchable sequences involving MOVW
instructions that can more easily be manipulated to carry a 12-bit
immediate. Note that the non-LPAE ARM sequence is not updated: MOVW
may not be supported on non-LPAE platforms, and the sequence itself
can be updated more easily to apply the 12 bits of displacement.
For Thumb2, which has many more versions of opcodes, switch to a sequence
that can be patched by the same patching code for both versions, and use
asm constraints and S-suffixed opcodes to force narrow encodings to be
selected.
Suggested-by: Russell King <linux@armlinux.org.uk>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
arch/arm/include/asm/memory.h | 43 +++++++++++----
arch/arm/kernel/head.S | 57 +++++++++++++-------
2 files changed, 69 insertions(+), 31 deletions(-)
diff --git a/arch/arm/include/asm/memory.h b/arch/arm/include/asm/memory.h
index 4121662dea5a..7184a2540816 100644
--- a/arch/arm/include/asm/memory.h
+++ b/arch/arm/include/asm/memory.h
@@ -183,6 +183,7 @@ extern const void *__pv_table_begin, *__pv_table_end;
#define PHYS_OFFSET ((phys_addr_t)__pv_phys_pfn_offset << PAGE_SHIFT)
#define PHYS_PFN_OFFSET (__pv_phys_pfn_offset)
+#ifndef CONFIG_THUMB2_KERNEL
#define __pv_stub(from,to,instr) \
__asm__("@ __pv_stub\n" \
"1: " instr " %0, %1, %2\n" \
@@ -192,25 +193,46 @@ extern const void *__pv_table_begin, *__pv_table_end;
: "=r" (to) \
: "r" (from), "I" (__PV_BITS_31_24))
-#define __pv_stub_mov_hi(t) \
- __asm__ volatile("@ __pv_stub_mov\n" \
- "1: mov %R0, %1\n" \
+#define __pv_add_carry_stub(x, y) \
+ __asm__ volatile("@ __pv_add_carry_stub\n" \
+ "0: movw %R0, %2\n" \
+ "1: adds %Q0, %1, %R0, lsl #24\n" \
+ "2: mov %R0, %3\n" \
+ " adc %R0, %R0, #0\n" \
" .pushsection .pv_table,\"a\"\n" \
- " .long 1b - .\n" \
+ " .long 0b - ., 1b - ., 2b - .\n" \
" .popsection\n" \
- : "=r" (t) \
- : "I" (__PV_BITS_7_0))
+ : "=&r" (y) \
+ : "r" (x), "j" (0), "I" (__PV_BITS_7_0) \
+ : "cc")
+
+#else
+#define __pv_stub(from,to,instr) \
+ __asm__("@ __pv_stub\n" \
+ "0: movw %0, %2\n" \
+ " lsls %0, #24\n" \
+ " " instr "s %0, %1, %0\n" \
+ " .pushsection .pv_table,\"a\"\n" \
+ " .long 0b - .\n" \
+ " .popsection\n" \
+ : "=&l" (to) \
+ : "l" (from), "j" (0) \
+ : "cc")
#define __pv_add_carry_stub(x, y) \
__asm__ volatile("@ __pv_add_carry_stub\n" \
- "1: adds %Q0, %1, %2\n" \
+ "0: movw %R0, %2\n" \
+ " lsls %R0, #24\n" \
+ " adds %Q0, %1, %R0\n" \
+ "1: mvn %R0, #0\n" \
" adc %R0, %R0, #0\n" \
" .pushsection .pv_table,\"a\"\n" \
- " .long 1b - .\n" \
+ " .long 0b - ., 1b - .\n" \
" .popsection\n" \
- : "+r" (y) \
- : "r" (x), "I" (__PV_BITS_31_24) \
+ : "=&l" (y) \
+ : "l" (x), "j" (0) \
: "cc")
+#endif
static inline phys_addr_t __virt_to_phys_nodebug(unsigned long x)
{
@@ -219,7 +241,6 @@ static inline phys_addr_t __virt_to_phys_nodebug(unsigned long x)
if (sizeof(phys_addr_t) == 4) {
__pv_stub(x, t, "add");
} else {
- __pv_stub_mov_hi(t);
__pv_add_carry_stub(x, t);
}
return t;
diff --git a/arch/arm/kernel/head.S b/arch/arm/kernel/head.S
index d2bd3b258386..86cea608a5ea 100644
--- a/arch/arm/kernel/head.S
+++ b/arch/arm/kernel/head.S
@@ -639,43 +639,45 @@ __fixup_a_pv_table:
mov r6, r6, lsr #24
cmn r0, #1
#ifdef CONFIG_THUMB2_KERNEL
- moveq r0, #0x200000 @ set bit 21, mov to mvn instruction
- lsls r6, #24
- beq .Lnext
- clz r7, r6
- lsr r6, #24
- lsl r6, r7
- bic r6, #0x0080
- lsrs r7, #1
- orrcs r6, #0x0080
- orr r6, r6, r7, lsl #12
- orr r6, #0x4000
+ moveq r0, #0x200 @ bit 9, ADD to SUB instruction (T1 encoding)
b .Lnext
.Lloop: add r7, r4
add r4, #4
+#ifdef CONFIG_ARM_LPAE
+ ldrh ip, [r7]
+ARM_BE8(rev16 ip, ip)
+ tst ip, #0x200 @ MOVW has bit 9 set, MVN has it clear
+ bne 0f @ skip if MOVW
+ tst r0, #0x200 @ need to convert MVN to MOV ?
+ bne .Lnext
+ eor ip, ip, #0x20 @ flick bit #5
+ARM_BE8(rev16 ip, ip)
+ strh ip, [r7]
+ b .Lnext
+0:
+#endif
ldrh ip, [r7, #2]
ARM_BE8(rev16 ip, ip)
- tst ip, #0x4000
- and ip, #0x8f00
- orrne ip, r6 @ mask in offset bits 31-24
- orreq ip, r0 @ mask in offset bits 7-0
+ orr ip, r6 @ mask in offset bits 31-24
ARM_BE8(rev16 ip, ip)
strh ip, [r7, #2]
- bne .Lnext
- ldrh ip, [r7]
+ ldrh ip, [r7, #6]
ARM_BE8(rev16 ip, ip)
- bic ip, #0x20
- orr ip, ip, r0, lsr #16
+ eor ip, ip, r0
ARM_BE8(rev16 ip, ip)
- strh ip, [r7]
+ strh ip, [r7, #6]
#else
#ifdef CONFIG_CPU_ENDIAN_BE8
@ in BE8, we load data in BE, but instructions still in LE
+#define PV_BIT20 0x00001000
#define PV_BIT22 0x00004000
+#define PV_BIT23_22 0x0000c000
#define PV_IMM8_MASK 0xff000000
#define PV_ROT_MASK 0x000f0000
#else
+#define PV_BIT20 0x00100000
#define PV_BIT22 0x00400000
+#define PV_BIT23_22 0x00c00000
#define PV_IMM8_MASK 0x000000ff
#define PV_ROT_MASK 0xf00
#endif
@@ -683,11 +685,26 @@ ARM_BE8(rev16 ip, ip)
moveq r0, #PV_BIT22 @ set bit 22, mov to mvn instruction
b .Lnext
.Lloop: ldr ip, [r7, r4]
+#ifdef CONFIG_ARM_LPAE
+ tst ip, #PV_BIT23_22 @ MOVW has bit 23:22 clear, MOV/ADD/SUB have it set
+ARM_BE8(rev ip, ip)
+ orreq ip, ip, r6
+ARM_BE8(rev ip, ip)
+ beq 2f
+ tst ip, #PV_BIT20 @ ADDS has bit 20 set
+ beq 1f
+ tst r0, #PV_BIT22 @ check whether to invert bits 23:22 (ADD -> SUB)
+ beq .Lnext
+ eor ip, ip, #PV_BIT23_22
+ b 2f
+1:
+#endif
bic ip, ip, #PV_IMM8_MASK
tst ip, #PV_ROT_MASK @ check the rotation field
orrne ip, ip, r6 ARM_BE8(, lsl #24) @ mask in offset bits 31-24
biceq ip, ip, #PV_BIT22 @ clear bit 22
orreq ip, ip, r0 @ mask in offset bits 7-0
+2:
str ip, [r7, r4]
add r4, r4, #4
#endif
--
2.17.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [RFC/RFT PATCH 6/6] ARM: p2v: reduce p2v alignment requirement to 2 MiB
2020-09-18 10:30 [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB Ard Biesheuvel
` (4 preceding siblings ...)
2020-09-18 10:31 ` [RFC/RFT PATCH 5/6] ARM: p2v: switch to MOVW for Thumb2 and ARM/LPAE Ard Biesheuvel
@ 2020-09-18 10:31 ` Ard Biesheuvel
2020-09-18 17:25 ` [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment " Ard Biesheuvel
2020-09-19 23:49 ` Nicolas Pitre
7 siblings, 0 replies; 14+ messages in thread
From: Ard Biesheuvel @ 2020-09-18 10:31 UTC (permalink / raw)
To: linux-arm-kernel
Cc: linux-efi, Nicolas Pitre, Linus Walleij, Russell King,
Santosh Shilimkar, Zhen Lei, Ard Biesheuvel
Update the p2v patching code so we can deal with displacements that are
not a multiple of 16 MiB but of 2 MiB, to prevent wasting of up to 14 MiB
of physical RAM when running on a platform where the start of memory is
not correctly aligned.
For the ARM code path, this simply comes down to using two add/sub
instructions instead of one for the carryless version, and patching
each of them with the correct immediate depending on the rotation
field. For the LPAE calculation, it patches the MOVW instruction with
up to 12 bits of offset.
For the Thumb2 code path, patching more than 11 bits off displacement
is somewhat cumbersome, and given that 11 bits produce a minimum
alignment of 2 MiB, which is also the granularity for LPAE block
mappings, it makes sense to stick to 2 MiB for the new p2v requirement.
Suggested-by: Zhen Lei <thunder.leizhen@huawei.com>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
arch/arm/Kconfig | 2 +-
arch/arm/include/asm/memory.h | 15 +++++++-----
arch/arm/kernel/head.S | 24 +++++++++++++++-----
3 files changed, 28 insertions(+), 13 deletions(-)
diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index e00d94b16658..c4737a0e613b 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -240,7 +240,7 @@ config ARM_PATCH_PHYS_VIRT
kernel in system memory.
This can only be used with non-XIP MMU kernels where the base
- of physical memory is at a 16MB boundary.
+ of physical memory is at a 2MiB boundary.
Only disable this option if you know that you do not require
this feature (eg, building a kernel for a single machine) and
diff --git a/arch/arm/include/asm/memory.h b/arch/arm/include/asm/memory.h
index 7184a2540816..5da01e7f0d8a 100644
--- a/arch/arm/include/asm/memory.h
+++ b/arch/arm/include/asm/memory.h
@@ -173,6 +173,7 @@ extern unsigned long vectors_base;
* so that all we need to do is modify the 8-bit constant field.
*/
#define __PV_BITS_31_24 0x81000000
+#define __PV_BITS_23_16 0x810000
#define __PV_BITS_7_0 0x81
extern unsigned long __pv_phys_pfn_offset;
@@ -187,16 +188,18 @@ extern const void *__pv_table_begin, *__pv_table_end;
#define __pv_stub(from,to,instr) \
__asm__("@ __pv_stub\n" \
"1: " instr " %0, %1, %2\n" \
+ "2: " instr " %0, %0, %3\n" \
" .pushsection .pv_table,\"a\"\n" \
- " .long 1b - .\n" \
+ " .long 1b - ., 2b - .\n" \
" .popsection\n" \
- : "=r" (to) \
- : "r" (from), "I" (__PV_BITS_31_24))
+ : "=&r" (to) \
+ : "r" (from), "I" (__PV_BITS_31_24), \
+ "I"(__PV_BITS_23_16))
#define __pv_add_carry_stub(x, y) \
__asm__ volatile("@ __pv_add_carry_stub\n" \
"0: movw %R0, %2\n" \
- "1: adds %Q0, %1, %R0, lsl #24\n" \
+ "1: adds %Q0, %1, %R0, lsl #20\n" \
"2: mov %R0, %3\n" \
" adc %R0, %R0, #0\n" \
" .pushsection .pv_table,\"a\"\n" \
@@ -210,7 +213,7 @@ extern const void *__pv_table_begin, *__pv_table_end;
#define __pv_stub(from,to,instr) \
__asm__("@ __pv_stub\n" \
"0: movw %0, %2\n" \
- " lsls %0, #24\n" \
+ " lsls %0, #21\n" \
" " instr "s %0, %1, %0\n" \
" .pushsection .pv_table,\"a\"\n" \
" .long 0b - .\n" \
@@ -222,7 +225,7 @@ extern const void *__pv_table_begin, *__pv_table_end;
#define __pv_add_carry_stub(x, y) \
__asm__ volatile("@ __pv_add_carry_stub\n" \
"0: movw %R0, %2\n" \
- " lsls %R0, #24\n" \
+ " lsls %R0, #21\n" \
" adds %Q0, %1, %R0\n" \
"1: mvn %R0, #0\n" \
" adc %R0, %R0, #0\n" \
diff --git a/arch/arm/kernel/head.S b/arch/arm/kernel/head.S
index 86cea608a5ea..d08d506a0ccd 100644
--- a/arch/arm/kernel/head.S
+++ b/arch/arm/kernel/head.S
@@ -614,8 +614,8 @@ __fixup_pv_table:
mov r0, r8, lsr #PAGE_SHIFT @ convert to PFN
str r0, [r6] @ save computed PHYS_OFFSET to __pv_phys_pfn_offset
strcc ip, [r7, #HIGH_OFFSET] @ save to __pv_offset high bits
- mov r6, r3, lsr #24 @ constant for add/sub instructions
- teq r3, r6, lsl #24 @ must be 16MiB aligned
+ mov r6, r3, lsr #21 @ constant for add/sub instructions
+ teq r3, r6, lsl #21 @ must be 2MiB aligned
THUMB( it ne @ cross section branch )
bne __error
str r3, [r7, #LOW_OFFSET] @ save to __pv_offset low bits
@@ -636,10 +636,13 @@ __fixup_a_pv_table:
add r6, r6, r0
ldr r0, [r6, #HIGH_OFFSET] @ pv_offset high word
ldr r6, [r6, #LOW_OFFSET] @ pv_offset low word
- mov r6, r6, lsr #24
+ mov r6, r6, lsr #16
cmn r0, #1
#ifdef CONFIG_THUMB2_KERNEL
moveq r0, #0x200 @ bit 9, ADD to SUB instruction (T1 encoding)
+ mov r3, r6, lsr #13 @ isolate top 3 bits of displacement
+ ubfx r6, r6, #5, #8 @ put bits 28:21 into the imm8 field
+ bfi r6, r3, #12, #3 @ put bits 31:29 into the imm3 field
b .Lnext
.Lloop: add r7, r4
add r4, #4
@@ -658,7 +661,7 @@ ARM_BE8(rev16 ip, ip)
#endif
ldrh ip, [r7, #2]
ARM_BE8(rev16 ip, ip)
- orr ip, r6 @ mask in offset bits 31-24
+ orr ip, r6 @ mask in offset bits 31-21
ARM_BE8(rev16 ip, ip)
strh ip, [r7, #2]
ldrh ip, [r7, #6]
@@ -674,21 +677,26 @@ ARM_BE8(rev16 ip, ip)
#define PV_BIT23_22 0x0000c000
#define PV_IMM8_MASK 0xff000000
#define PV_ROT_MASK 0x000f0000
+#define PV_ROT16_MASK 0x00080000
#else
#define PV_BIT20 0x00100000
#define PV_BIT22 0x00400000
#define PV_BIT23_22 0x00c00000
#define PV_IMM8_MASK 0x000000ff
#define PV_ROT_MASK 0xf00
+#define PV_ROT16_MASK 0x800
#endif
moveq r0, #PV_BIT22 @ set bit 22, mov to mvn instruction
+ and r3, r6, #0xf
+ mov r6, r6, lsr #8
b .Lnext
.Lloop: ldr ip, [r7, r4]
#ifdef CONFIG_ARM_LPAE
tst ip, #PV_BIT23_22 @ MOVW has bit 23:22 clear, MOV/ADD/SUB have it set
ARM_BE8(rev ip, ip)
- orreq ip, ip, r6
+ orreq ip, ip, r6, lsl #4
+ orreq ip, ip, r3, lsr #4
ARM_BE8(rev ip, ip)
beq 2f
tst ip, #PV_BIT20 @ ADDS has bit 20 set
@@ -701,9 +709,13 @@ ARM_BE8(rev ip, ip)
#endif
bic ip, ip, #PV_IMM8_MASK
tst ip, #PV_ROT_MASK @ check the rotation field
- orrne ip, ip, r6 ARM_BE8(, lsl #24) @ mask in offset bits 31-24
biceq ip, ip, #PV_BIT22 @ clear bit 22
orreq ip, ip, r0 @ mask in offset bits 7-0
+ beq 2f
+
+ tst ip, #PV_ROT16_MASK @ amount of shift?
+ orreq ip, ip, r6 ARM_BE8(, lsl #24) @ mask in offset bits 31-24
+ orrne ip, ip, r3 ARM_BE8(, lsl #24) @ mask in offset bits 23-16
2:
str ip, [r7, r4]
add r4, r4, #4
--
2.17.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 14+ messages in thread
* Re: [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB
2020-09-18 10:30 [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB Ard Biesheuvel
` (5 preceding siblings ...)
2020-09-18 10:31 ` [RFC/RFT PATCH 6/6] ARM: p2v: reduce p2v alignment requirement to 2 MiB Ard Biesheuvel
@ 2020-09-18 17:25 ` Ard Biesheuvel
2020-09-19 23:49 ` Nicolas Pitre
7 siblings, 0 replies; 14+ messages in thread
From: Ard Biesheuvel @ 2020-09-18 17:25 UTC (permalink / raw)
To: Linux ARM
Cc: linux-efi, Nicolas Pitre, Linus Walleij, Russell King,
Santosh Shilimkar, Zhen Lei
On Fri, 18 Sep 2020 at 12:31, Ard Biesheuvel <ardb@kernel.org> wrote:
>
> This series is inspired by Zhei Len's series [0], which updates the
> ARM p2v patching code to optionally support p2v relative alignments
> of as little as 64 KiB.
>
> Reducing this alignment is necessary for some specific Huawei boards,
> but given that reducing this minimum alignment will make the boot
> sequence more robust for all platforms, especially EFI boot, which
> no longer relies on the 128 MB masking of the decompressor load address,
> but uses firmware memory allocation routines to find a suitable spot
> for the decompressed kernel.
>
> This series is not based on Zhei Len's code, but addresses the same
> problem, and takes some feedback given in the review into account:
> - use of a MOVW instruction to avoid two adds/adcs sequences when dealing
> with the carry on LPAE
> - add support for Thumb2 kernels as well
> - make the change unconditional - it will bit rot otherwise, and has value
> for other platforms as well.
>
> The first four patches are general cleanup and preparatory changes.
> Patch #5 implements the switch to a MOVW instruction without changing
> the minimum alignment.
> Patch #6 reduces the minimum alignment to 2 MiB.
>
> Tested on QEMU in ARM/!LPAE, ARM/LPAE, Thumb2/!LPAE and Thumb2/LPAE modes.
>
> Cc: Zhen Lei <thunder.leizhen@huawei.com>
> Cc: Russell King <rmk+kernel@armlinux.org.uk>
> Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
> Cc: Linus Walleij <linus.walleij@linaro.org>
> Cc: Nicolas Pitre <nico@fluxnic.net>
>
> [0] https://lore.kernel.org/linux-arm-kernel/20200915015204.2971-1-thunder.leizhen@huawei.com/
>
> Ard Biesheuvel (6):
> ARM: p2v: factor out shared loop processing
> ARM: p2v: factor out BE8 handling
> ARM: p2v: drop redundant 'type' argument from __pv_stub
> ARM: p2v: use relative references in patch site arrays
> ARM: p2v: switch to MOVW for Thumb2 and ARM/LPAE
> ARM: p2v: reduce p2v alignment requirement to 2 MiB
>
Note: there is a thinko in this version of the patches, as it
unnecessarily patches ADD instructions into SUB and vice versa. Since
QEMU has its DRAM at 0x4000_0000, the translation for a 3g/1g split is
0x8000_0000, which why it worked in my testing, but will fail on other
platforms (or with other splits, for that matter)
For the general discussion of the approach taken here, that does not
really matter, so I won't send a v2 until people have had some time to
have a look. However, if you want to test this code on other hardware,
please use the code at the following link instead:
https://git.kernel.org/pub/scm/linux/kernel/git/ardb/linux.git/log/?h=arm-p2v-v2
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB
2020-09-18 10:30 [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB Ard Biesheuvel
` (6 preceding siblings ...)
2020-09-18 17:25 ` [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment " Ard Biesheuvel
@ 2020-09-19 23:49 ` Nicolas Pitre
2020-09-20 7:50 ` Ard Biesheuvel
2020-09-20 8:55 ` Russell King - ARM Linux admin
7 siblings, 2 replies; 14+ messages in thread
From: Nicolas Pitre @ 2020-09-19 23:49 UTC (permalink / raw)
To: Ard Biesheuvel
Cc: linux-efi, Linus Walleij, Russell King, Santosh Shilimkar,
Zhen Lei, linux-arm-kernel
On Fri, 18 Sep 2020, Ard Biesheuvel wrote:
> This series is inspired by Zhei Len's series [0], which updates the
> ARM p2v patching code to optionally support p2v relative alignments
> of as little as 64 KiB.
>
> Reducing this alignment is necessary for some specific Huawei boards,
> but given that reducing this minimum alignment will make the boot
> sequence more robust for all platforms, especially EFI boot, which
> no longer relies on the 128 MB masking of the decompressor load address,
> but uses firmware memory allocation routines to find a suitable spot
> for the decompressed kernel.
>
> This series is not based on Zhei Len's code, but addresses the same
> problem, and takes some feedback given in the review into account:
> - use of a MOVW instruction to avoid two adds/adcs sequences when dealing
> with the carry on LPAE
> - add support for Thumb2 kernels as well
> - make the change unconditional - it will bit rot otherwise, and has value
> for other platforms as well.
>
> The first four patches are general cleanup and preparatory changes.
> Patch #5 implements the switch to a MOVW instruction without changing
> the minimum alignment.
> Patch #6 reduces the minimum alignment to 2 MiB.
>
> Tested on QEMU in ARM/!LPAE, ARM/LPAE, Thumb2/!LPAE and Thumb2/LPAE modes.
At this point I think this really ought to be split into a file of its
own... and maybe even rewritten in C. Even though I wrote the original
code, I no longer understand it without re-investing time into it. But
in either cases the whole of head.S would need to have its registers
shuffled first to move long lived values away from r0-r3,ip,lr to allow
for standard function calls.
Nicolas
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB
2020-09-19 23:49 ` Nicolas Pitre
@ 2020-09-20 7:50 ` Ard Biesheuvel
2020-09-20 8:57 ` Russell King - ARM Linux admin
2020-09-20 8:55 ` Russell King - ARM Linux admin
1 sibling, 1 reply; 14+ messages in thread
From: Ard Biesheuvel @ 2020-09-20 7:50 UTC (permalink / raw)
To: Nicolas Pitre
Cc: linux-efi, Linus Walleij, Russell King, Santosh Shilimkar,
Zhen Lei, Linux ARM
On Sun, 20 Sep 2020 at 01:49, Nicolas Pitre <nico@fluxnic.net> wrote:
>
> On Fri, 18 Sep 2020, Ard Biesheuvel wrote:
>
> > This series is inspired by Zhei Len's series [0], which updates the
> > ARM p2v patching code to optionally support p2v relative alignments
> > of as little as 64 KiB.
> >
> > Reducing this alignment is necessary for some specific Huawei boards,
> > but given that reducing this minimum alignment will make the boot
> > sequence more robust for all platforms, especially EFI boot, which
> > no longer relies on the 128 MB masking of the decompressor load address,
> > but uses firmware memory allocation routines to find a suitable spot
> > for the decompressed kernel.
> >
> > This series is not based on Zhei Len's code, but addresses the same
> > problem, and takes some feedback given in the review into account:
> > - use of a MOVW instruction to avoid two adds/adcs sequences when dealing
> > with the carry on LPAE
> > - add support for Thumb2 kernels as well
> > - make the change unconditional - it will bit rot otherwise, and has value
> > for other platforms as well.
> >
> > The first four patches are general cleanup and preparatory changes.
> > Patch #5 implements the switch to a MOVW instruction without changing
> > the minimum alignment.
> > Patch #6 reduces the minimum alignment to 2 MiB.
> >
> > Tested on QEMU in ARM/!LPAE, ARM/LPAE, Thumb2/!LPAE and Thumb2/LPAE modes.
>
> At this point I think this really ought to be split into a file of its
> own... and maybe even rewritten in C. Even though I wrote the original
> code, I no longer understand it without re-investing time into it. But
> in either cases the whole of head.S would need to have its registers
> shuffled first to move long lived values away from r0-r3,ip,lr to allow
> for standard function calls.
>
I agree with that in principle, however, running C code with a stack
with the MMU off is slightly risky.
I have managed to simplify the code a bit more (given that some
patching was not needed to begin with), and I can add some more
comments to head.S to annotate the actions.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB
2020-09-19 23:49 ` Nicolas Pitre
2020-09-20 7:50 ` Ard Biesheuvel
@ 2020-09-20 8:55 ` Russell King - ARM Linux admin
1 sibling, 0 replies; 14+ messages in thread
From: Russell King - ARM Linux admin @ 2020-09-20 8:55 UTC (permalink / raw)
To: Nicolas Pitre
Cc: linux-efi, Linus Walleij, Santosh Shilimkar, Zhen Lei,
Ard Biesheuvel, linux-arm-kernel
On Sat, Sep 19, 2020 at 07:49:26PM -0400, Nicolas Pitre wrote:
> On Fri, 18 Sep 2020, Ard Biesheuvel wrote:
> > This series is inspired by Zhei Len's series [0], which updates the
> > ARM p2v patching code to optionally support p2v relative alignments
> > of as little as 64 KiB.
> >
> > Reducing this alignment is necessary for some specific Huawei boards,
> > but given that reducing this minimum alignment will make the boot
> > sequence more robust for all platforms, especially EFI boot, which
> > no longer relies on the 128 MB masking of the decompressor load address,
> > but uses firmware memory allocation routines to find a suitable spot
> > for the decompressed kernel.
> >
> > This series is not based on Zhei Len's code, but addresses the same
> > problem, and takes some feedback given in the review into account:
> > - use of a MOVW instruction to avoid two adds/adcs sequences when dealing
> > with the carry on LPAE
> > - add support for Thumb2 kernels as well
> > - make the change unconditional - it will bit rot otherwise, and has value
> > for other platforms as well.
> >
> > The first four patches are general cleanup and preparatory changes.
> > Patch #5 implements the switch to a MOVW instruction without changing
> > the minimum alignment.
> > Patch #6 reduces the minimum alignment to 2 MiB.
> >
> > Tested on QEMU in ARM/!LPAE, ARM/LPAE, Thumb2/!LPAE and Thumb2/LPAE modes.
>
> At this point I think this really ought to be split into a file of its
> own... and maybe even rewritten in C. Even though I wrote the original
> code, I no longer understand it without re-investing time into it. But
> in either cases the whole of head.S would need to have its registers
> shuffled first to move long lived values away from r0-r3,ip,lr to allow
> for standard function calls.
However, that code has to run _before_ the virtual mappings are setup,
which makes C code out of the question, unless we build a separate
executable binary and then insert it into the kernel image. So, sorry,
it's not going to be practical to rewrite it in C.
--
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB
2020-09-20 7:50 ` Ard Biesheuvel
@ 2020-09-20 8:57 ` Russell King - ARM Linux admin
2020-09-20 10:06 ` Ard Biesheuvel
0 siblings, 1 reply; 14+ messages in thread
From: Russell King - ARM Linux admin @ 2020-09-20 8:57 UTC (permalink / raw)
To: Ard Biesheuvel
Cc: linux-efi, Nicolas Pitre, Linus Walleij, Santosh Shilimkar,
Zhen Lei, Linux ARM
On Sun, Sep 20, 2020 at 09:50:30AM +0200, Ard Biesheuvel wrote:
> On Sun, 20 Sep 2020 at 01:49, Nicolas Pitre <nico@fluxnic.net> wrote:
> >
> > On Fri, 18 Sep 2020, Ard Biesheuvel wrote:
> >
> > > This series is inspired by Zhei Len's series [0], which updates the
> > > ARM p2v patching code to optionally support p2v relative alignments
> > > of as little as 64 KiB.
> > >
> > > Reducing this alignment is necessary for some specific Huawei boards,
> > > but given that reducing this minimum alignment will make the boot
> > > sequence more robust for all platforms, especially EFI boot, which
> > > no longer relies on the 128 MB masking of the decompressor load address,
> > > but uses firmware memory allocation routines to find a suitable spot
> > > for the decompressed kernel.
> > >
> > > This series is not based on Zhei Len's code, but addresses the same
> > > problem, and takes some feedback given in the review into account:
> > > - use of a MOVW instruction to avoid two adds/adcs sequences when dealing
> > > with the carry on LPAE
> > > - add support for Thumb2 kernels as well
> > > - make the change unconditional - it will bit rot otherwise, and has value
> > > for other platforms as well.
> > >
> > > The first four patches are general cleanup and preparatory changes.
> > > Patch #5 implements the switch to a MOVW instruction without changing
> > > the minimum alignment.
> > > Patch #6 reduces the minimum alignment to 2 MiB.
> > >
> > > Tested on QEMU in ARM/!LPAE, ARM/LPAE, Thumb2/!LPAE and Thumb2/LPAE modes.
> >
> > At this point I think this really ought to be split into a file of its
> > own... and maybe even rewritten in C. Even though I wrote the original
> > code, I no longer understand it without re-investing time into it. But
> > in either cases the whole of head.S would need to have its registers
> > shuffled first to move long lived values away from r0-r3,ip,lr to allow
> > for standard function calls.
> >
>
> I agree with that in principle, however, running C code with a stack
> with the MMU off is slightly risky.
It's more than "slightly". C code has literal addresses, which are raw
virtual addresses. These are meaningless with the MMU off.
I guess one could correct the various pointers the code would read, but
you could not directly access any variable (as that involves
dereferencing a virtual address stored in the function's literal pool.)
--
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB
2020-09-20 8:57 ` Russell King - ARM Linux admin
@ 2020-09-20 10:06 ` Ard Biesheuvel
2020-09-20 15:34 ` Nicolas Pitre
0 siblings, 1 reply; 14+ messages in thread
From: Ard Biesheuvel @ 2020-09-20 10:06 UTC (permalink / raw)
To: Russell King - ARM Linux admin
Cc: linux-efi, Nicolas Pitre, Linus Walleij, Santosh Shilimkar,
Zhen Lei, Linux ARM
On Sun, 20 Sep 2020 at 10:57, Russell King - ARM Linux admin
<linux@armlinux.org.uk> wrote:
>
> On Sun, Sep 20, 2020 at 09:50:30AM +0200, Ard Biesheuvel wrote:
> > On Sun, 20 Sep 2020 at 01:49, Nicolas Pitre <nico@fluxnic.net> wrote:
> > >
> > > On Fri, 18 Sep 2020, Ard Biesheuvel wrote:
> > >
> > > > This series is inspired by Zhei Len's series [0], which updates the
> > > > ARM p2v patching code to optionally support p2v relative alignments
> > > > of as little as 64 KiB.
> > > >
> > > > Reducing this alignment is necessary for some specific Huawei boards,
> > > > but given that reducing this minimum alignment will make the boot
> > > > sequence more robust for all platforms, especially EFI boot, which
> > > > no longer relies on the 128 MB masking of the decompressor load address,
> > > > but uses firmware memory allocation routines to find a suitable spot
> > > > for the decompressed kernel.
> > > >
> > > > This series is not based on Zhei Len's code, but addresses the same
> > > > problem, and takes some feedback given in the review into account:
> > > > - use of a MOVW instruction to avoid two adds/adcs sequences when dealing
> > > > with the carry on LPAE
> > > > - add support for Thumb2 kernels as well
> > > > - make the change unconditional - it will bit rot otherwise, and has value
> > > > for other platforms as well.
> > > >
> > > > The first four patches are general cleanup and preparatory changes.
> > > > Patch #5 implements the switch to a MOVW instruction without changing
> > > > the minimum alignment.
> > > > Patch #6 reduces the minimum alignment to 2 MiB.
> > > >
> > > > Tested on QEMU in ARM/!LPAE, ARM/LPAE, Thumb2/!LPAE and Thumb2/LPAE modes.
> > >
> > > At this point I think this really ought to be split into a file of its
> > > own... and maybe even rewritten in C. Even though I wrote the original
> > > code, I no longer understand it without re-investing time into it. But
> > > in either cases the whole of head.S would need to have its registers
> > > shuffled first to move long lived values away from r0-r3,ip,lr to allow
> > > for standard function calls.
> > >
> >
> > I agree with that in principle, however, running C code with a stack
> > with the MMU off is slightly risky.
>
> It's more than "slightly". C code has literal addresses, which are raw
> virtual addresses. These are meaningless with the MMU off.
>
> I guess one could correct the various pointers the code would read, but
> you could not directly access any variable (as that involves
> dereferencing a virtual address stored in the function's literal pool.)
>
We might be able to work around that by compiling with -fPIC, and/or
by ensuring that all inputs to the routine are passed via function
parameters. But I agree that using C for this code is probably not the
right choice.
If there is no disagreement about the 2 MiB alignment, or the choice
of opcodes for the patchable sequences, I can prepare a v2 that fixes
the issues I mentioned, and has some more explanatory comments in the
patching routine.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB
2020-09-20 10:06 ` Ard Biesheuvel
@ 2020-09-20 15:34 ` Nicolas Pitre
0 siblings, 0 replies; 14+ messages in thread
From: Nicolas Pitre @ 2020-09-20 15:34 UTC (permalink / raw)
To: Ard Biesheuvel
Cc: linux-efi, Linus Walleij, Russell King - ARM Linux admin,
Santosh Shilimkar, Zhen Lei, Linux ARM
On Sun, 20 Sep 2020, Ard Biesheuvel wrote:
> On Sun, 20 Sep 2020 at 10:57, Russell King - ARM Linux admin
> <linux@armlinux.org.uk> wrote:
> >
> > On Sun, Sep 20, 2020 at 09:50:30AM +0200, Ard Biesheuvel wrote:
> > > On Sun, 20 Sep 2020 at 01:49, Nicolas Pitre <nico@fluxnic.net> wrote:
> > > >
> > > > On Fri, 18 Sep 2020, Ard Biesheuvel wrote:
> > > >
> > > > > This series is inspired by Zhei Len's series [0], which updates the
> > > > > ARM p2v patching code to optionally support p2v relative alignments
> > > > > of as little as 64 KiB.
> > > > >
> > > > > Reducing this alignment is necessary for some specific Huawei boards,
> > > > > but given that reducing this minimum alignment will make the boot
> > > > > sequence more robust for all platforms, especially EFI boot, which
> > > > > no longer relies on the 128 MB masking of the decompressor load address,
> > > > > but uses firmware memory allocation routines to find a suitable spot
> > > > > for the decompressed kernel.
> > > > >
> > > > > This series is not based on Zhei Len's code, but addresses the same
> > > > > problem, and takes some feedback given in the review into account:
> > > > > - use of a MOVW instruction to avoid two adds/adcs sequences when dealing
> > > > > with the carry on LPAE
> > > > > - add support for Thumb2 kernels as well
> > > > > - make the change unconditional - it will bit rot otherwise, and has value
> > > > > for other platforms as well.
> > > > >
> > > > > The first four patches are general cleanup and preparatory changes.
> > > > > Patch #5 implements the switch to a MOVW instruction without changing
> > > > > the minimum alignment.
> > > > > Patch #6 reduces the minimum alignment to 2 MiB.
> > > > >
> > > > > Tested on QEMU in ARM/!LPAE, ARM/LPAE, Thumb2/!LPAE and Thumb2/LPAE modes.
> > > >
> > > > At this point I think this really ought to be split into a file of its
> > > > own... and maybe even rewritten in C. Even though I wrote the original
> > > > code, I no longer understand it without re-investing time into it. But
> > > > in either cases the whole of head.S would need to have its registers
> > > > shuffled first to move long lived values away from r0-r3,ip,lr to allow
> > > > for standard function calls.
> > > >
> > >
> > > I agree with that in principle, however, running C code with a stack
> > > with the MMU off is slightly risky.
> >
> > It's more than "slightly". C code has literal addresses, which are raw
> > virtual addresses. These are meaningless with the MMU off.
> >
> > I guess one could correct the various pointers the code would read, but
> > you could not directly access any variable (as that involves
> > dereferencing a virtual address stored in the function's literal pool.)
> >
>
> We might be able to work around that by compiling with -fPIC, and/or
> by ensuring that all inputs to the routine are passed via function
> parameters. But I agree that using C for this code is probably not the
> right choice.
Yeah... It is doable, like we do in the decompressor case, but the level
of caution needed here would probably negate the gain from writing this
in C.
The argument for moving it out to a file of its own still stands though.
> If there is no disagreement about the 2 MiB alignment, or the choice
> of opcodes for the patchable sequences, I can prepare a v2 that fixes
> the issues I mentioned, and has some more explanatory comments in the
> patching routine.
Yes please. Given you do have it all in your head now, it would be very
valuable to be way more expensive with comments. Adding a comment block
with the opcode bit definition before the code that transforms them
would also be nice.
Nicolas
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 14+ messages in thread
end of thread, other threads:[~2020-09-20 15:36 UTC | newest]
Thread overview: 14+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-09-18 10:30 [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment to 2 MiB Ard Biesheuvel
2020-09-18 10:30 ` [RFC/RFT PATCH 1/6] ARM: p2v: factor out shared loop processing Ard Biesheuvel
2020-09-18 10:30 ` [RFC/RFT PATCH 2/6] ARM: p2v: factor out BE8 handling Ard Biesheuvel
2020-09-18 10:30 ` [RFC/RFT PATCH 3/6] ARM: p2v: drop redundant 'type' argument from __pv_stub Ard Biesheuvel
2020-09-18 10:31 ` [RFC/RFT PATCH 4/6] ARM: p2v: use relative references in patch site arrays Ard Biesheuvel
2020-09-18 10:31 ` [RFC/RFT PATCH 5/6] ARM: p2v: switch to MOVW for Thumb2 and ARM/LPAE Ard Biesheuvel
2020-09-18 10:31 ` [RFC/RFT PATCH 6/6] ARM: p2v: reduce p2v alignment requirement to 2 MiB Ard Biesheuvel
2020-09-18 17:25 ` [RFC/RFT PATCH 0/6] ARM: p2v: reduce min alignment " Ard Biesheuvel
2020-09-19 23:49 ` Nicolas Pitre
2020-09-20 7:50 ` Ard Biesheuvel
2020-09-20 8:57 ` Russell King - ARM Linux admin
2020-09-20 10:06 ` Ard Biesheuvel
2020-09-20 15:34 ` Nicolas Pitre
2020-09-20 8:55 ` Russell King - ARM Linux admin
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).